melogit problems with "initial values not feasible"

[Cindy Lin]

Dear Statalists,
I encounter the "initial values not feasible" problem when running multi-level logit regression.
I tried to simplify the model (null model) but have still encountered the same problem.

Also tried to increase iterate numbles and use technique(nr), but it still encountered the "initial values not feasible" problem.

When I tried to use technique(bhhh) it works, but cannot estimate the AIC and BIC.

Are there any other ways we can try to solve this problem?

All, the best,

[Rich Goldstein]

given the lack of information about your data and your question of interest, I can't help re: the "initial values" issue;

however, your last issue is that you used a non-existent command (or made a typo); the command should be:

estat ic

for more on this, see

help melogit postestimation